Insect Visitation and Pollen Deposition in an Invaded Prairie Plant Community

Home , Linum, Panula (moth)

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by DigitalCommons@University of Nebraska

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

USGS Northern Prairie Wildlife Research Center US Geological Survey

2006

Insect Visitation and Pollen Deposition in an Invaded Prairie Plant Community

Diane L. Larson USGS Northern Prairie Wildlife Research Center, [email protected]

Ronald A. Royer Minot State University

Margaret R. Royer Minot State University

Follow this and additional works at: https://digitalcommons.unl.edu/usgsnpwrc

Part of the Other International and Area Studies Commons

Larson, Diane L.; Royer, Ronald A.; and Royer, Margaret R., "Insect Visitation and Pollen Deposition in an Invaded Prairie Plant Community" (2006). USGS Northern Prairie Wildlife Research Center. 85. https://digitalcommons.unl.edu/usgsnpwrc/85

This Article is brought to you for free and open access by the US Geological Survey at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in USGS Northern Prairie Wildlife Research Center by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. BIOLOGICAL CONSERVATION 130 (2006) 148– 159

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/biocon

Insect visitation and pollen deposition in an invaded prairie plant community

Diane L. Larsona,*, Ronald A. Royerb, Margaret R. Royerb aUSGS Northern Prairie Wildlife Research Center, 100 Ecology Building, 1987 Upper Buford Circle, St. Paul, MN 55108, United States bDivision of Science, Minot State University, Minot, ND, United States

ARTICLE INFO ABSTRACT

Article history: Invasive plants with large flowering displays have been shown to compete with native Received 15 July 2005 plants for pollinator services, often to the detriment of native plant fitness. In this study, Received in revised form we compare the pollinator communities and pollen deposited on stigmas of native plant 28 November 2005 species within and away from stands of the invasive alien plant, leafy spurge (Euphorbia Accepted 14 December 2005 esula) at a large natural area in North Dakota, USA. Specifically, we ask if infestation influ- Available online 26 January 2006 ences (1) visitation rates and taxonomic composition of visitors to native flowers, and (2) the amount of conspecific pollen, number of pollen species, and proportion of heterospe- Keywords: cific pollen on stigmas of native plants. We observed visits to selected native species during Biological invasion May and June 2000 and 2001. Stigmas were collected from a subsample of the flowers Euphorbia esula within these plots, squashed, and the pollen identified and counted under a light micro- Halictidae scope. Visitation varied between years and among species of native plants: infestation Mixed-grass prairie had mixed effects in 2000 but visitation, especially by halictids was always lower within Pollen on stigmas infestations in 2001. Despite differences in visitation between years, we found significantly less conspecific pollen on stigmas from infested plots in six of eight cases; we never found significantly more conspecific pollen on stigmas from within infestations. Our results emphasize the temporal variability in plant–pollinator relations and the added complexity imposed by an invasive species that will always make prediction of effects difficult. Nonetheless, the consistently lower conspecific pollen counts on native stigmas within infestations, regardless of visitation, suggest the likelihood of negative effects. 2005 Elsevier Ltd. All rights reserved.

1. Introduction infestation could be enhanced (Feldman et al., 2004); however, if those pollinators preferentially visit the invasive species, Many alien plants form dense monospecific stands that visitation to, and therefore pollination of, native plants could flower profusely, and thus may introduce substantial be depressed (Rathcke, 1983). Because pollination systems amounts of alien pollen and nectar into the native communi- tend to be generalized, especially in the northern temperate ties they invade. The diffuse nature of pollination mutualisms zone of North America (Johnson and Steiner, 2000), pollina- (Palmer et al., 2003) and the multiple interspecific interactions tion of native plants also could be depressed if pollinators involved (Waser et al., 1996) make the outcome of such an in- that visit native plants deposit alien pollen on their stigmas, flux of resources highly uncertain. If an increase in resources and much of the native pollen is carried to aliens, rather than attracts a disproportionately larger number of pollinators to to conspecifics (improper pollen transfer (IPT); Waser, 1978; the area, pollination of native plants in the vicinity of the Rathcke, 1983). Alien pollen that makes its way to native plant

* Corresponding author: Tel.: +1 612 625 9271; fax: +1 612 624 6777. E-mail address: [email protected] (D.L. Larson). 0006-3207/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2005.12.009 BIOLOGICAL CONSERVATION 130 (2006) 148– 159 149

stigmas may affect reproduction by clogging or mechanically Vegetation at the park is dominated by native cool season blocking the stigma or style, by chemically interfering with grasses, primarily Pascopyrum smithii and Stipa spp. The south fertilization (allelopathy), or by producing hybrids (Brown unit of the park was established in 1947 and contains free- and Mitchell, 2001; McLernon et al., 1996; Waser, 1978). ranging native ungulates as well as a small herd of wild Indeed, evidence has begun to accumulate that suggests horses. More detail on plant species associations within the invasive plants with large flowering displays can compete park can be found in Larson et al. (2001). E. esula was first doc- with native plants for pollinator services, often to the detri- umented in the South Unit of TRNP in 1970 (unpublished park ment of native plant fitness. Effects have been documented documents). Infested and non-infested areas used in this in terms of both pollen quantity and quality (Brown and study had been so since at least 1996. Mitchell, 2001; Brown et al., 2002; Chittka and Schu¨ rkens, Euphorbia esula is a Eurasian perennial that invades mixed- 2001; Moragues and Traveset, 2005). What is not yet clear is grass prairie in rangeland and natural areas throughout the whether the change in pollination success is due to the northern Great Plains and causes extensive environmental change in the prevalence of exotic pollen, which is dispersed and economic damage (Bangsund et al., 1996, 1999; Trammell (at the expense of native pollen) by the same generalist polli- and Butler, 1995). Aerial photography in 1991 indicated nator community that existed prior to infestation, a change in approximately 1620 ha of Euphorbia in the South Unit of TRNP, the pollinator community that visits native plants to fewer or with the distribution closely aligned with watercourses and less efficient pollinators, a change in rates of visitation to drainages. Euphorbia flowers profusely during June in North flowers, or some combination of these. If plants compete for Dakota, bearing terminal umbels with sequentially maturing pollinator services, a new plant species that provides copious female and male flowers (cyathia); each ‘‘wave’’ of flowers resources will dramatically alter the competitive relation- produces 5–12 female followed by 11–20 male flowers and ships within the plant community, with less abundant spe- each stem may branch to produce as many as 16 umbels (Sel- cies being put at a distinct disadvantage (Palmer et al., leck et al., 1962). Each plant’s yellow flowers mature in succes- 2003), especially with respect to constant pollinators that spe- sion over a period of several weeks. Pollen remains viable for cialize on abundant and rewarding flowers (e.g., Gegear and up to 2 days on each flower and nectar is produced by the Laverty, 1998, 2004; Kunin, 1993). In contrast, if pollinators glands of the cyathium 3–8 days after inversion of the female compete for floral rewards, introduction of a super-abundant flower (Selleck et al., 1962), so a consistent source of pollen resource may mean that pollen and nectar are no longer lim- and nectar remains available for approximately 6 weeks at iting, and new or formerly out-competed insect taxa may en- our study sites. Capsules, which are explosively dehiscent, ter the system. These issues are of concern not only with mature in mid- to late-summer and contain up to three seeds. respect to potential effects on pollen-limited plants, but also in light of recent indications of declining pollinator popula- 2.2. Sampling design tions (Kearns et al., 1998; Spira, 2001). In this study, we compare the pollinator insect communi- Using data from a 1996 stratified-random survey of plants in ties and pollen delivered to stigmas of native plant species the South Unit in which >800 geo-referenced transects were within and >100 m away from stands of the invasive alien examined (methods for the survey can be found in Larson plant, leafy spurge (Euphorbia esula L.; Euphorbiaceae; hereaf- et al. (2001)), we identified areas within the South Unit that ter Euphorbia). Specifically, we ask (1) does visitation to native were likely to have one of several concurrently blooming na- flowers by pollinator insect taxa vary between infested and tive plants, either >100 m away from or within a Euphorbia non-infested sites, (2) does the amount of conspecific pollen infestation. We avoided the dwarf sagebrush (Artemesia cana) on stigmas vary between infested and non-infested sites, and river bottom vegetation types because they harbor large and (3) does infestation influence the number of pollen species populations of a variety of non-native species (Larson et al., and proportion of heterospecific pollen on native plant stig- 2001). All sampling was done in upland prairie to minimize mas? If the presence of Euphorbia facilitates pollination of bias in insect habitat characteristics that could be associated native species, we would expect more visitation and greater with infested versus non-infested habitats. For an area to be amounts of conspecific pollen on stigmas of native flowers in considered infested, it had to have flowering Euphorbia stems infested sites compared with non-infested sites. On the other intermixed with flowering native species; total area of each hand, if Euphorbia competes with native species for pollinators, infestation varied from tens of square meters to a hectare we would expect fewer visits and less conspecific pollen on na- or more. In consideration of the necessary sample size, only tive stigmas in infested sites. Less conspecific pollen on native relatively common species were selected as target native spe- stigmas could also result from a change in the pollinator com- cies. We narrowed our target native species to only those that, munity within infested sites to species that are less constant in upon field inspection, occurred both within and at least 100 m visitation habit; if this is the case, we would expect to find a away from an infestation and that bloomed concurrently with higher diversity of pollen species on stigmas in infested sites. Euphorbia. While 100 m is admittedly arbitrary, it was the largest area we could reasonably search exhaustively for the 2. Methods presence of Euphorbia. Our goal was to minimize the chance that insects that had visited Euphorbia would arrive at our 2.1. Study site non-infested plots. Target native species in 2000 were Cam- panula rotundifolia L. (Campanulaceae) for visitation only and The study was conducted in the South Unit of Theodore Roo- Linum lewisii Pursh var. lewisii (Linaceae) and Oxytropis sevelt National Park (TRNP) in western North Dakota, USA. lambertii Pursh var. lambertii (Fabaceae) for pollen counts and 150 BIOLOGICAL CONSERVATION 130 (2006) 148– 159

visitation. In 2001 we added Calylophus serrulatus (Nutt.) Raven 2.4. Pollen loads on stigmas (Onagraceae), Sphaeralcea coccinea (Nutt.) Rydb. ssp. coccinea (Malvaceae), and Vicia americana Muhl. ex Willd. (Fabaceae) Prior to beginning observations (described above), four flow- for pollen counts (we were unable to obtain a sufficient sam- ers of the target species in each plot were chosen arbitrarily ple size to analyze visitation to these species), and conducted and marked with a small bit of pastel-colored thread, each a pollen counts on Campanula. Of these, Linum, Campanula, Cal- different color. The thread did not appear to influence pollin- ylophus and Sphaeralcea as well as Euphorbia have actinomor- ators; none investigated the thread and marked flowers were phic (i.e., radially symmetrical) flowers, although Campanula no more likely to be visited than unmarked flowers (data not is bell-shaped and must be entered from below, and Oxytropis shown). After observations were completed for that day, we and Vicia have zygomorphic (i.e., bilaterally symmetrical). collected stigmas of the four marked flowers in each plot Taxonomy follows the USDA Plants data base (USDA 2004). and placed them individually in glassine envelopes. Upon We use the genus name to refer to these species throughout returning to the lab, stigmas were placed on microscope the paper. slides with a drop of resin-based mounting medium. The cover slip was pressed firmly over the stigma to squash it 2.3. Visitation and force the pollen into a single layer, suitable for counting (Kearns and Inouye, 1993). We assessed visitation to native flowers and to those of We made a reference pollen collection of all flowering Euphorbia within 1 m2 square plots that contained at least plants in the vicinity of our plots. We did not collect grass pol- six (total) individual flowering stems of one or more of the na- len, however, which may account for some of the pollen tive target species. Plots were located in the vicinity of one of grains we were unable to identify. Prior to counting pollen the survey transects (described above) where a target species on collected stigmas, counters familiarized themselves with occurred and were positioned so that the required number of the reference collection by hand drawing and measuring pol- target species was included in the plot. We counted the num- len from each collected species. Pollen on stigmas was identi- ber of flowers of each target species, Euphorbia inflorescences, fied and counted by visually scanning the entire slide, since and any other species in flower within the plot. In addition, squashing the stigma spread the pollen out away from it. we counted the number of flowers by species in six When a pollen grain could not be identified using the refer- 0.25 · 0.25 m quadrats in each of the four cardinal compass ence collection or other resources available to us, it was given directions outside the plot, since flower density may influence a unique ID that we used to classify other grains on other visitation and our plots might not adequately reflect density slides. Each of three counters kept their own list of un- at the scale perceived by pollinators. Visitation on most plots knowns, which we did not try to reconcile, so the number was assessed once in the morning and once mid-day, weather of unknown species is inflated. Stigmas from 653 flowers were permitting; results were averaged for each plot prior to anal- evaluated for pollen loads. In all, we counted 175,507 pollen ysis. Different plots were established each day. grains divided among 23 identified pollen species and 6192 Observations lasted 20 min, during which time each polli- pollen grains divided among 38 unknown pollen morpho- nation event on each target flower was recorded on a data types; approximately 3% of pollen grains counted were of un- sheet. We defined a pollination event (hereafter termed known species. ‘‘visit’’) as any contact of an insect with anthers or stigmas on an individual flower. We recorded the flower species vis- 2.5. Statistical analysis ited and the insect to the lowest field-identifiable taxonomic category (see below). At the end of the 20-min period, a new We used mixed models with type III sums of squares (PROC plot was established in a new location and the process re- MIXED; SAS 8.02 (SAS Institute, 1999)) to identify differences peated. Observations were made between 07:00 and 14:30 in visitation rates to native plant species in plots in Euphor- (most after 08:00) with temperatures between 14 and 35 C. bia-infested sites and in sites >100 m from an infestation. We did not conduct observations if winds were in excess of Dependent variables were number of visits to flowers of na- Beaufort 4 (5–6 knots) or if it was raining. We assessed insect tive species, which were analyzed separately; independent visitation on 233 plots (77.67 h at 20 min/plot) over the two variables included visitor taxon, infestation, year and all pos- year period and recorded 763 individual insect visits in 2000 sible interactions. Plot was the random variable nested within and 812 in 2001. infestation. A parallel analysis was conducted with genus A haphazard sample of bees was collected at each area within the Halictidae replacing visitor taxon as an indepen- where visitation was assessed during the 2000 field season dent variable for 2001 only. The abundance of flowering con- and sent to a taxonomic expert for identification. Observers specifics and other plant species within and adjacent to the were trained to identify common bee families prior to begin- visitation plot were used as covariates to control for differ- ning work. They were often able to distinguish Bombus, Apis, ences in flower abundance among plots; the number of indi- Halictus, Dialictus, Agapostemon and Lasioglossum, but native viduals of other species was never statistically significant bees could not be identified to species in the field. We did and was excluded from analyses presented in this paper. not distinguish among taxa of wasps because we could not We also used mixed models to determine the effect of reliably discern taxa in the field, or lepidopterans because vis- infestation on (1) number of conspecific pollen grains, itation rates were low enough that individual taxa could not (2) number of leafy spurge pollen grains, (3) number of pollen be analyzed statistically. Flies were overwhelmingly muscoid, species, and (4) proportion of heterospecific pollen on stigmas and taxa were not separated for analysis. of each native species. Because we had two years of data for BIOLOGICAL CONSERVATION 130 (2006) 148– 159 151

Linum and Oxytropis, we ran one analysis with infestation, Halictidae. This analysis was restricted to 2001, when we had stigma species, year and all possible interactions as indepen- more confidence in our identification of the bees. Halictids in dent variables for these two species. A second analysis, for the four genera above comprised 66% of our observations of 2001 only, used all species sampled in that year, with infesta- bees in 2001, with the remainder made up of apids (Apis melli- tion, stigma species and their interaction as independent fera, the only exotic bee encountered, was recorded in only six variables. Total number of conspecific flowers within and plots), andrenids, megachilids and unidentified halictids adjacent to the plot was used as a covariate to separate effects (based on the sample identified in 2000; see Section 2); trends of flower abundance from effects of infestation on conspecific in the unknown bees collectively followed those of the halict- pollen. All quantitative variables with the exception of pro- ids. Overall, we recorded nearly three times as many visits by portion of heterospecific pollen were log transformed (after halictids in non-infested sites as we did in infested sites adding 1) to improve normality. Proportion of heterospecific (0.96 ± 0.08 and 0.33 ± 0.06, ln-transformed least square pollen was square-root transformed prior to analysis, also mean ± SE, respectively) in 2001. Visits to Linum varied with to improve normality. both bee taxon and infestation, but there was no interaction between the variables (Fig. 2 and Table 2). There were fewer 3. Results visits by halictids of all taxa within infested sites than in non-infested sites. Dialictus and Halictus were the most com- Plots contained 1.1 ± 0.04 flowering species (mean ± SE) in mon genera of bees to visit Linum. Visits by halictids to Cam- non-infested areas and 1.9 ± 0.03 (mean ± SE) in infested panula were also significantly reduced in sites infested by areas (F = 303.87, df = 1, 427, p < 0.0001); the higher number Euphorbia. Dialictus spp. tended to be the most common visi- of species within infestations is attributable to the presence tors to Campanula, but the differences among bee taxa were of Euphorbia in plots. Of 299 observations of insects moving not statistically significant. Oxytropis had a non-significant between flowers within a plot, only 8 involved >1 species. trend toward higher visitation by halictids in non-infested The number of conspecific target flowers (other than leafy sites, but there were no differences among taxa. Euphorbia it- spurge) in a plot averaged 15.4 ± 0.86 in non-infested areas self was more commonly visited by Dialictus spp. than other and 11.0 ± 0.65 in infested areas, a significant difference bee taxa. (F = 16.41, df = 1, 427, p < 0.0001); total flowers in the plots (including Euphorbia) did not vary between non-infested and 3.2. Pollen loads on stigmas infested plots (15.4 ± 1.1 and 17.5 ± 0.8, respectively, F = 2.36, df = 1, 427, p = .1252). Plots in infested areas averaged The amount of conspecific pollen on stigmas of Linum and 6.4 ± 7.0 leafy spurge stems. Oxytropis was greater in the absence of Euphorbia than within an infestation (Table 3), and this did not vary between 2000 3.1. Visitation and 2001 (Table 4). Very little Euphorbia pollen was found on these two native species and this did not vary significantly Native species in infested plots received 89% as many visits as with infestation or species. Suggestive of preferences of gen- in non-infested plots in 2000. In contrast, in 2001 native spe- eralist pollinators for Linum, we found more pollen species cies in infested plots received only 57% as many total visits as and a larger proportion of heterospecific pollen on Linum than did those in non-infested plots (F = 3.93, df = 1, 82, p = 0.05 for on Oxytropis stigmas (Table 3), and this was not influenced by the infestation by year interaction). Significantly more bees infestation and did not vary between 2000 and 2001 (Table 4). visited Linum in non-infested than infested sites in 2001, but Looking only at 2001, but including all six species sampled not 2000 when visitation was similar between infestation lev- that year, similar patterns emerge. The presence of Euphorbia els (Fig. 1 and Table 1; least square means and standard errors within 100 m was significantly associated with fewer conspe- for all visitation analyses are included in Appendix A). Bees cific pollen grains on stigmas of native species (Fig. 3 and Ta- and flies were the most common visitors to Linum. Like Linum, ble 5). The small amount of Euphorbia pollen we found on Campanula received more visits by bees in non-infested than native stigmas was marginally greater in infested than in in infested areas in 2001, but the reverse was true in 2000, non-infested sites (Fig. 3 and Table 5). Two species, Calylophus when more visits were recorded in infested than in non-in- and Vicia, had no Euphorbia pollen at all and could not be infested areas (Fig. 1). Flies were more common visitors to all cluded in the analysis. The fact that we encountered Euphor- native plant species in 2001 than in 2000. We recorded no vis- bia pollen in non-infested sites implies that the pollen its by Lepidoptera or wasps to Campanula, and few by Bombus. moved >100 m, at least in a few cases. Overall, 65 of the 653 Unlike Linum and Campanula, infestation by Euphorbia did not flowers examined had Euphorbia pollen on at least one of their significantly influence visitation to Oxytropis (Fig. 1). The sig- stigmas; 13 of these were collected in non-infested sites and nificant insect taxon by year interaction (Table 1) reflected 42 in infested sites. We found fewer pollen species on stigmas strong shifts in visitation by Bombus, flies and Lepidoptera of native plants in infested than in non-infested sites but the to Oxytropis between years. In contrast to the native species, proportion of heterospecific pollen on stigmas was greater for we found no significant differences in visitation to Euphorbia most species in infested than in non-infested sites (Fig. 3). Li- itself between years (Table 1); flies were by far the most com- num was an exception to this trend, with a larger proportion mon visitors in both years (Fig. 1). of heterospecific pollen in non-infested sites in 2001. Campan- We also examined visitation by native bee taxa, other than ula and Sphaeralcea were most likely to have several pollen Bombus, we could identify in the field, including the genera species on their stigmas while Vicia never and Oxytropis very Dialictus, Halictus, Agapostemon and Lasioglossum in the family rarely had pollen from other species on their stigmas. 152 BIOLOGICAL CONSERVATION 130 (2006) 148– 159

a. Visits by pollinators 2000-2001

2.5 Campanula Linum Oxytropis Euphorbia

2.0 Bees Bombus Diptera 1.5 Lepidoptera Wasps

1.0 Ln ((visits/20 min)+1)

0.5

0.0 Infestation NYY N NY NY N Y NY Y Y 2000 2001 2000 2001 2000 2001 2000 2001

b. Visits by halictid bees, 2001 2.5 Campanula Linum Oxytropis Euphorbia

2.0

Halictid sp Dialictus 1.5 Halictus Lasioglossum Agapostemon 1.0

0.5 Ln((visits/20 min)+1)

0.0

-0.5 No Yes No Yes No Yes Yes Infestation

Fig. 1 – Cumulative visitation by pollinator taxa to native species and Euphorbia. (a) Visitation by all taxa in 2000–2001; (b) visitation by halictid bees in 2001.

4. Discussion Moragues and Traveset (2005) examined effects of the invasive Carpobrotus on several native species on an island in 4.1. Visitation and conspecific pollen the Mediterranean and found similar variability in effects between years. Although, like Moragues and Traveset (2005),we The relationship between Euphorbia infestation and pollina- found greater levels of visitation in infested plots for one spe- tion of native plants at our study sites was not as straight-for- cies (Campanula) in one year, the lack of increased conspecific ward as our hypotheses implied it would be. Although both pollen on stigmas of Campanula suggests that we did not ob- visitation and deposition of conspecific pollen were lower serve facilitation. Our results show that effects of infestation on infested plots in most cases, this varied both by plant spe- on insect visitation exist, but do not give an indication of how cies and between the two years of the study. Year-to-year var- consistently they are expressed. Nonetheless, we never found iation in pollination studies is to be expected: the variability significantly greater amounts of conspecific pollen on stigmas inherent in pollination systems is well documented and from infested areas and, in six of eight cases, the amount of many have warned of the difficulties in interpreting short- conspecific pollen was significantly less on stigmas from in- term studies (Fishbein and Venable, 1996; Herrera, 1988, fested areas. 1995; Price et al., 2005; Roubik, 2001). Although we found clear Visitation and the amount of conspecific pollen on stigmas effects of infestation on visitation by bees in 2001, results were not correlated in our study. Mean amount of conspecific were more variable for 2000, when visits to Campanula were pollen on Linum stigmas from non-infested plots did not vary higher in infested than non-infested sites and visits to Oxytro- between years even though overall visitation was much higher pis and Linum did not differ significantly with infestation. on non-infested plots in 2001 than 2000. Further, despite BIOLOGICAL CONSERVATION 130 (2006) 148– 159 153

Table 1 – ANOVA results for visitation by insects to Linum lewisii, Campanula rotundifolia, Oxytropis lambertii and Euphorbia esula in 2000–2001 Target species Effect Num. DF Den. DF F value Pr > F

L. lewisii Infestation 1 100 4.95 0.0283 Insect taxa 4 404 56.72 <0.0001 Year 1 404 1.92 0.1663 Infestation · taxa 4 404 1.24 0.2938 Infestation · year 1 404 1.62 0.2043 Taxa · year 4 404 2.15 0.0741 Infestation · taxa · year 4 404 3.75 0.0053 Covariate 1 404 16.23 <0.0001

C. rotundifolia Infestation 1 33 0.21 0.6522 Insect taxa 4 131 24.84 <0.0001 Year 1 131 0.19 0.6664 Infestation · taxa 4 131 0.05 0.9947 Infestation · year 1 131 6.21 0.0139 Taxa · year 4 131 2.01 0.0962 Infestation · taxa · year 4 131 6.18 0.0001 Covariate 1 131 0 0.9594

O. lambertii Infestation 1 42 0.8 0.3763 Insect taxa 4 172 3.26 0.0132 Year 1 172 0.01 0.9362 Infestation · taxa 4 172 0.87 0.4808 Infestation · year 1 172 3.3 0.0709 Taxa · year 4 172 2.55 0.0411 Infestation · taxa · year 4 172 1.22 0.3032 Covariate 1 172 9.48 0.0024

E. esula Year 1 348 0.95 0.3311 Insect taxa 4 348 34.76 <0.0001 Year · taxa 4 348 0.95 0.4353 Covariate 1 348 16.3 <0.0001

The covariate is the number of ﬂowers of the target species within the plot.

similar visitation rates between infested and non-infested found that alien plants were mainly pollinated by general- plots in 2000 for both Linum and Oxytropis, conspecific pollen ists. Results of our study agree, in that flies were the pri- numbers were still lower on stigmas from infested plots. mary visitors to Euphorbia. Although bees were observed These observations imply that not only does the quantity of to visit Euphorbia, the magnitude of this visitation did not, pollinators vary among years, but perhaps more importantly, on average, balance the deficit seen in visitation to natives so does the quality: more visitation does not always mean in infested sites. more conspecific pollen. Lau and Galloway (2004), studying Flies did make a substantial contribution to the pollinator the effects of low-efficiency, or ‘‘ugly’’, pollinators on male community of one native species, Linum, and they have been function in Campanula americana, cautioned that increased vis- shown to effectively pollinate this species in other locations itation rates cannot be assumed to translate directly to in- (Kearns and Inouye, 1994). Flies were marginally less common creased fitness. visitors to Linum in infested than non-infested plots in 2000, Having ruled out facilitation, does Euphorbia compete leaving open the possibility that Euphorbia was attracting poll- with native species for pollinators? Although lower levels inators from Linum in this year. Note that in 2001, when flies of visitation and less conspecific pollen within infestations visited all species relatively frequently, there was no effect of suggest competition, the observational nature of this study infestation on fly visits to Linum. Thus, effects of infestation and our level of taxonomic resolution do not allow a defin- likely depend on the population size of pollinators in a given itive answer. However, the prevalence and consistency of year. Diptera in the pollinator assemblage of Euphorbia, compared Visitation by pollinators is known to be strongly influ- with the prevalence of bees in that of the native species, enced by flower density (Kunin, 1997). We statistically ac- suggests that, unlike the congeners Lythrum salicaria and L. counted for differences in native flower density between alatum (Brown and Mitchell, 2001), these native plants infested and non-infested plots in this study, yet still found may be experiencing relatively little competition for pollina- a significant effect of infestation on conspecific pollen depo- tor services with the invasive. Memmott and Waser (2002) sition. The native species we targeted were common and demonstrated via food-web analyses that invasive plant pollen was still sufficiently abundant to fertilize all the species have become integrated into the interaction webs ovules (but see Mitchell (1997)). However, the combined ef- of the pollinator communities they studied. However, they fects of low density and less conspecific pollen transported 154 BIOLOGICAL CONSERVATION 130 (2006) 148– 159

a. Conspecific pollen grains/stigma b. E. esula pollen grains/stigma 7 0.3

6 0.2

0.1 4

3 0.0

-0.1 Ln((mean pollen grains/stigma)+1) Ln((mean pollen grains/stigma)+1) 1

0 -0.2 2000 20012000 2001 2000 20012000 2001 Non infested Linum Oxytropis Linum Oxytropis Infested

c. Number of pollen species/stigma d. Proportion heterospecific pollen/stigma

1.4 0.4

1.2 0.3

1.0 0.2

0.8

0.1 0.6

0.0 0.4

0.2 -0.1 Ln((mean pollen species/stigma)+1) Sqrt(proportion nonconspecific pollen/stigma) 0.0 -0 .2 2000 20012000 2001 2000 20012000 2001 Linum Oxytropis Linum Oxytropis

Fig. 2 – Pollen counts for stigmas collected from Linum lewisii and Oxytropis lambertii in 2000 and 2001. Shown are least square means of transformed (ln(n + 1)) counts; proportions were square-root transformed. (a) Conspeciﬁc pollen; (b) Euphorbia pollen; (c) number of pollen species; (d) proportion of heterospeciﬁc pollen.

Table 2 – ANOVA results for visitation by halictid bees to Linum lewisii, Campanula rotundifolia, Oxytropis lambertii and Euphorbia esula in 2001 Target species Effect Num. DF Den. DF F value Pr > F

L. lewisii Infested 1 58 8.85 0.0043 Insect taxa 4 231 19.5 <.0001 Infested · taxa 4 231 2.47 0.0452 Covariate 1 231 6.89 0.0092

C. rotundifolia Infested 1 20 4.43 0.0481 Insect taxa 4 84 3.95 0.0055 Infested · taxa 4 84 2.12 0.0856 Covariate 1 84 1.06 0.3052

O. lambertii Infested 1 13 5.75 0.0322 Insect taxa 4 56 1.96 0.1126 Infested · taxa 4 56 1.96 0.1126 Covariate 1 56 5.48 0.0228

E. esula Insect taxa 4 236 2.66 0.0333 Covariate 1 236 0.01 0.9292

The covariate is the number of ﬂowers of the target species within the plot. BIOLOGICAL CONSERVATION 130 (2006) 148– 159 155

Table 3 – Effects of infestation and native species from which the stigma was collected (stigma species) on pollen deposition on native stigmas

Variable Effect Infestation Stigma sp. Meana Std. error

Conspeciﬁc pollen No 4.3 0.16 Yes 3.9 0.14 Linum 2.6 0.095 Oxytropis 5.5 0.19

Number of pollen species Linum 0.99 0.024 Oxytropis 0.68 0.052

Proportion of heterospeciﬁc pollen Linum 0.22 0.020 Oxytropis 0.0048 0.043

Least square means and their standard errors are shown for the highest order effect with a significant (p 6 0.05) F value (see Table 4). Note that there were no significant effects of Euphorbia pollen on stigmas, so that variable is not included in this table. a Units for ‘‘Conspecific pollen’’ are Ln(n + 1) pollen grains/stigma; units for ‘‘Number of pollen species’’ are Ln(n +1) N species; units for p ‘‘Proportion heterospecific pollen’’ are proportion of total pollen on stigma that was not conspecific.

Table 4 – ANOVA results for pollen counts on stigmas of Linum lewisii and Oxytropis lambertii (listed as ‘‘Stigma species’’) in 2000 and 2001 Variable Type 3 tests of ﬁxed effects Effect Num. DF Den. DF FP

Conspeciﬁc pollen Infestation 1 101 4.85 0.0299 Stigma sp. 1 305 187.19 <.0001 Year 1 305 3.42 0.0652 Infestation · stigma sp. 1 305 0.09 0.7696 Infestation · year 1 305 0.01 0.9243 Stigma sp. · year 1 305 3.03 0.0825 Infestation · stigma sp. · year 1 305 0.74 0.3919 Covariate 1 305 0.69 0.407

Leafy spurge pollen Infestation 1 342 1.59 0.2087 Stigma sp. 1 342 2.86 0.0919 Year 1 342 1.59 0.2086 Infestation · stigma sp. 1 342 3.41 0.0656 Infestation · year 1 342 3.33 0.0689 Stigma sp. · year 1 342 3.41 0.0658 Infestation · stigma sp. · year 1 342 1.53 0.2169

Number of pollen species Infestation 1 342 0.11 0.7382 Stigma sp. 1 342 28.39 <.0001 Year 1 342 0.13 0.7238 Infestation · stigma sp. 1 342 0.41 0.52 Infestation · year 1 342 2.17 0.1419 Stigma sp. · year 1 342 1.54 0.2155 Infestation · stigma sp. · year 1 342 2.89 0.0899

Proportion heterospeciﬁc pollen Infestation 1 332 0.01 0.9225 Stigma sp. 1 332 21.17 <.0001 Year 1 332 0.03 0.8542 Infestation · stigma sp. 1 332 0 0.9954 Infestation · year 1 332 3.09 0.0795 Stigma sp. · year 1 332 0.15 0.7006 Infestation · stigma sp. · year 1 332 3.42 0.0652

The covariate is the number of ﬂowers of the target species within and adjacent to the plot.

in infested areas would likely put rare species within Euphor- of competition described above. There may have been quali- bia infestations at risk of pollen limitation, at least in some ties that varied systematically between infested and non-in- years. fested areas that were important to pollinators but that we The observational nature of this study imposes some lim- did not measure, such as nest sites for ground-dwelling bees, itations on our interpretation, in addition to the assessment or suitable woody vegetation for stem-nesters. 156 BIOLOGICAL CONSERVATION 130 (2006) 148– 159

a. Mean conspecific pollen grains on stigmas, 2001 b. Mean leafy spurge pollen on stigmas, 2001

10 0.5

8 0.4

6 0.3

4 0.2 Ln(n pollen grains+1) Ln(n pollen grains + 1)

2 0.1

0 0.0 us ula um Campanula Linum Oxytropis Sphaeralcea yloph pan Lin Vicea Cal Cam OxytropisSphaeralcea Stigma species Non infested Stigma species Infested c. Number of pollen species/stigma, 2001 d. Proportion heterospecific pollen/stigma, 2001

1.6 0.6

1.4 0.5

1.2

0.4 1.0

0.8 0.3

0.6 0.2 Sqrt(proportion pollen) Ln(n pollen species+1) 0.4

0.1 0.2

0.0 0.0 us ula um pis cea ea us ula um pis cea ea yloph pan Lin xytro eral Vic yloph pan Lin xytro eral Vic Cal Cam O Spha Cal Cam O Spha Stigma species Stigma species

Fig. 3 – Pollen counts for stigmas collected from Calylophus, Campanula, Linum, Oxytropis, Sphaeralcea and Vicia in 2001. Shown are least square means of transformed (ln(n + 1)) counts; proportions were square-root transformed. (a) Conspeciﬁc pollen; (b) Euphorbia pollen; (c) number of pollen species; (d) proportion of heterospeciﬁc pollen.

Table 5 – ANOVA results for counts of pollen on stigmas of Calylophus serrulatus, Campanula rotundifolia, Linum lewisii, Oxytropis lambertii, Sphaeralcea coccinea and Vicia americana in 2001 Response variable Type 3 tests of ﬁxed effects Effect Num. DF Den. DF FP

Conspeciﬁc pollen Infestation 1 92 4.78 0.0314 Stigma sp. 5 275 81.88 <.0001 Infestation · stigma sp. 5 275 1.48 0.1963 Covariate 1 275 0.72 0.3961

Leafy spurge pollen Infestation 1 257 3.99 0.0468 Stigma sp. 3 257 8.14 <.0001 Infestation · stigma sp. 3 257 2.94 0.0336

Number of pollen species Infestation 1 291 7.37 0.007 Stigma sp. 5 291 7.46 <.0001 Infestation · stigma sp. 5 291 1.18 0.3171

Proportion heterospeciﬁc pollen Infestation 1 280 0.17 0.684 Stigma sp. 5 280 3.73 0.0027 Infestation · stigma sp. 5 280 2.46 0.0332 BIOLOGICAL CONSERVATION 130 (2006) 148– 159 157

4.2. Improper pollen transfer proposition. Approximately 10% of the flowers we sampled had Euphorbia pollen. The amount of Euphorbia pollen on stig- In addition to influencing visitation rates and deposition of mas varied only slightly between infested and non-infested conspecific pollen, native plants within infestations may sites, thus indicating that the invasive pollen is transported, experience increased deposition of heterospecific pollen. at least occasionally, >100 m, while interspecific competition Empirical studies have generally found little invasive pollen for light and nutrients typically occurs within a few cm of on native stigmas (Aigner, 2004; Moragues and Traveset, the plant. 2005) and our study is no exception. We found only very small amounts of Euphorbia pollen on native stigmas (no 4.3. Management implications species had a back-transformed mean >0.5 grains/stigma), although the amounts varied significantly among native spe- The primary effects of Euphorbia infestation on pollination of cies; Campanula and Linum had greater amounts of Euphorbia native plants were a decline in conspecific pollen deposited pollen than other species in at least one year of the study. on stigmas in both years and in visitation by bees, especially Zygomorphic species Oxytropis and Vicia were largely im- Halictus spp. and Dialictus spp. in the family Halictidae, in mune from Euphorbia pollen, as was actinomorphic species one of the two years studied. Given that we detected de- Calylophus: none was ever found on Vicia or Calylophus stig- clines in conspecific pollen in species that are relatively mas. Although pollinators may be more likely to act as gen- common, the potential for significant pollen shortfalls in eralists among morphologically similar species (Kunin, 1993, rare species is high. Managers need to be aware of the dan- 1997), we found no evidence that species more similar to ger of pollen limitation, especially in rare or uncommon Euphorbia, in terms of flower shape (e.g., Linum, Sphaeralcia) plant species. Clonal or long-lived species may persist for or color (Calylophus) received more Euphorbia pollen than considerable periods of time in the absence of effective pol- did dissimilar flowers (e.g., Campanula, Oxytropis, Vicia). Poll- lination, leading to what Johnson and Steiner (2000) have inators may discern patterns of flower similarities that differ called ‘‘an insidious form of delayed extinction’’. Most land from our broad categorizations of symmetry and visible managers are aware of the locations of species of concern color, however. on their properties. We would encourage them to monitor If Euphorbia infestations attract greater numbers of gener- seed set (which is considerably easier to accomplish than alist pollinators, we would expect higher numbers of pollen are studies of pollination) of the insect-pollinated species if species and greater proportions of heterospecific pollen on significant encroachment of invasive species is noted. If native species within these infestations. In general, we found reductions in seed set are noted, this should trigger more fewer species of pollen on stigmas from infested sites and extensive studies of potential mechanisms, including pollen inconsistent differences in proportions of heterospecific pol- limitation and IPT. len, suggesting that, if Euphorbia is attracting generalist poll- Of equal importance is the observation that native halictid inators, this does not translate into IPT for native species. bees are less frequent visitors in infested areas in some years. Although contrary to our expectation, this result may reflect Despite a general increase in visits by halictids in non- an indirect effect of leafy spurge infestation, in that species infested areas between 2000 and 2001, visits in infested areas richness has been found to be lower within dense infesta- were down substantially over the same time period. Our re- tions (Belcher (1989); Larson, unpublished data). If fewer sults add to recent concern over the general decline in pollin- plant species exist within the foraging area of pollinators, ators (Kearns et al., 1998) and accentuate the need to fewer pollen species would be available for deposition on understand the role that invasive species play at all levels of stigmas. the ecosystems they invade. Because halictids are notoriously The variability we found in heterospecific pollen on difficult to identify, they present a special challenge to land stigmas once again emphasizes our limited ability to gener- managers who wish to monitor their populations. Additional alize about effects of infestation. Given the relatively low research to determine the reasons for fewer halictids within amounts of Euphorbia pollen and proportions of hetero- Euphorbia infestations might suggest indirect methods of specific pollen (ranging from a back-transformed mean of assessing risk to halictid populations that would be more <1% for Vicia and Oxytropis to 14–18% for Sphaeralcea), the accessible to land managers than direct monitoring of the differences we found may have little biological significance bee populations. for most species. However, we did find that introduction of Finally, the variability we observed between years requires Euphorbia pollen onto hand pollinated Linum flowers re- further investigation. We found no obvious correlations with sulted in significantly lower seed set than that found for local differences in temperature or precipitation between flowers on the same plant that were only hand pollinated the two years, so we cannot say what might have caused (least square means = 5 ± 0.84 and 9 ± 0.64 for spurge intro- the variability we observed. Managers will need to know duced vs. no spurge on fully pollinated flowers, F = 7.34, how frequently to expect effects of infestation to occur and df = 2, 8, p = 0.0155; Larson, unpublished data), leaving open to know if concurrency with other environmental fluctuations the possibility that small quantities of Euphorbia pollen may might exacerbate effects of infestation. have fitness effects. Chittka and Schu¨ rkens (2001) proposed that competition Acknowledgments between native and invasive plants may occur at substantially greater distances than previously thought, due to the We gratefully acknowledge field assistance provided by A. distances traveled by pollinators. Our results support that Haas, G. Hanley, D. Lang, J. Larson, J. Plummer, P. Scherr and 158 BIOLOGICAL CONSERVATION 130 (2006) 148– 159

L. Van Riper and pollen counting by J. Larson and K. Jacob- Target Infested Insect Year Visits/ Std. son. D. Yanega identiﬁed voucher specimens of bees. We are species taxa 20 min error grateful for comments on a previous draft by S. Huerd, N. Jor- dan, R. Mitchell, C. Reed, N. Waser, two anonymous reviewers, no Diptera 2001 0.17 0.11 and members of the Jordan and Larson lab groups, that signif- yes Diptera 2001 0.29 0.12 icantly improved the manuscript. We thank staff at Theodore no Lepidoptera 2000 0.00 0.20 Roosevelt National Park who provided logistical support for yes Lepidoptera 2000 0.00 0.12 this study and for many others over the years. Funding was no Lepidoptera 2001 0.00 0.11 provided by the US Geological Survey’s Park Oriented Biologi- yes Lepidoptera 2001 0.00 0.12 cal Support Program and Northern Prairie Wildlife Research no Wasps 2000 0.00 0.20 Center. yes Wasps 2000 0.00 0.12 no Wasps 2001 0.00 0.11 Appendix A yes Wasps 2001 0.00 0.12

O. lambertii no Bees 2000 0.43 0.08 Visitation by insects to Linum lewisii, Campanula rotundifolia, yes Bees 2001 0.30 0.11 Oxytropis lambertii and Euphorbia esula. Part a includes all in- no Bombus 2000 0.18 0.08 sect visitor taxa across the two years of the study. Note that yes Bombus 2001 0.09 0.11 ‘‘Bees’’ exclude bumblebees, which are listed as ‘‘Bombus’’. no Diptera 2000 0.03 0.08 Part b includes only those bees in the family Halictidae in yes Diptera 2001 0.33 0.11 2001. Least square means (ln(n + 1) pollen grains/stigma) no Lepidoptera 2000 0.27 0.08 and their standard errors are shown for the highest order ef- yes Lepidoptera 2001 0.05 0.11 fect with a signiﬁcant (p 6 0.05) F value. no Wasps 2000 0.01 0.08 yes Wasps 2001 0.09 0.11

E. esula yes Bees 0.22 0.05 Target Infested Insect Year Visits/20 Std. yes Bombus 0.00 0.05 species taxa min error yes Diptera 0.73 0.05 Part a yes Lepidoptera 0.01 0.05 L. lewisii no Bees 2000 0.74 0.10 yes Wasps 0.16 0.05

yes Bees 2000 0.83 0.11 Target Infested Insect Visits/20 Std. no Bees 2001 1.22 0.08 species taxa min error yes Bees 2001 0.59 0.09 no Bombus 2000 0.02 0.10 Part b yes Bombus 2000 0.07 0.11 L. lewisii No Agapostemon 0.08 0.10 no Bombus 2001 0.04 0.08 Yes Agapostemon 0.01 0.11 yes Bombus 2001 0.03 0.09 No Dialictus 0.71 0.10 no Diptera 2000 0.40 0.10 Yes Dialictus 0.29 0.11 yes Diptera 2000 0.20 0.11 No Halictid sp. 1.01 0.10 no Diptera 2001 0.55 0.08 Yes Halictid sp. 0.45 0.11 yes Diptera 2001 0.58 0.09 No Halictus 0.33 0.10 no Lepidoptera 2000 0.18 0.10 Yes Halictus 0.06 0.11 yes Lepidoptera 2000 0.05 0.11 No Lasioglossum 0.13 0.10 no Lepidoptera 2001 0.08 0.08 Yes Lasioglossum 0.06 0.11 yes Lepidoptera 2001 0.01 0.09 C. rotundifolia No 0.39 0.10 no Wasps 2000 0.03 0.10 Yes 0.07 0.11 yes Wasps 2000 0.00 0.11 Agapostemon 0.10 0.11 no Wasps 2001 0.06 0.08 Dialictus 0.36 0.11 yes Wasps 2001 0.03 0.09 Halictid sp. 0.48 0.11 C. rotundifolia no Bees 2000 0.56 0.20 Halictus 0.14 0.11 yes Bees 2000 1.38 0.12 Lasioglossum 0.06 0.11 no Bees 2001 1.00 0.11 O. lambertii No 0.15 0.04 yes Bees 2001 0.27 0.12 Yes 0.00 0.04 no Bombus 2000 0.00 0.20 yes Bombus 2000 0.11 0.12 E. esula Yes Agapostemon 0.07 0.04 no Bombus 2001 0.06 0.11 Yes Dialictus 0.10 0.04 yes Bombus 2001 0.00 0.12 Yes Halictid sp. 0.13 0.04 no Diptera 2000 0.00 0.20 Yes Halictus 0.02 0.04 yes Diptera 2000 0.04 0.12 Yes Lasioglossum 0.00 0.04 BIOLOGICAL CONSERVATION 130 (2006) 148– 159 159

REFERENCES Kunin, W.E., 1993. Sex and the single mustard: population density and pollinator behavior effects on seed-set. Ecology 74, 2145–2160. Kunin, W.E., 1997. Population size and density effects in Aigner, P.A., 2004. Ecological and genetic effects on demographic pollination – pollinator foraging and plant reproductive processes: pollination, clonality and seed production in success in experimental arrays of Brassica kaber. Journal of Dithyrea maritima. Biological Conservation 116, 27–34. Ecology 85, 225–234. Bangsund, D.A., Leitch, J.A., Leistritz, F.L., 1996. Economics of Larson, D.L., Anderson, P.J., Newton, W., 2001. Alien plant invasion herbicide control of leafy spurge (Euphorbia esula L). Journal of in mixed-grass prairie: effects of vegetation type and Agricultural and Resource Economics 21, 381–395. anthropogenic disturbance. Ecological Applications 11, Bangsund, D.A., Leistritz, F.L., Leitch, J.A., 1999. Assessing 128–141. economic impacts of biological control of weeds: The case Lau, J.A., Galloway, L.F., 2004. Effects of low-efficiency pollinators of leafy spurge in the northern Great Plains of the United on plant fitness and floral trait evolution in Campanula States. Journal of Environmental Management 56, americana (Campanulaceae). Oecologia 141, 577–583. 35–43. McLernon, S.M., Murphy, S.D., Aarssen, L.W., 1996. Heterospecific Belcher, J.W., Wilson, S.D., 1989. Leafy spurge and the species pollen transfer between sympatric species in a composition of a mixed-grass prairie. Journal of Range midsuccessional old-field community. American Journal of Management 42, 172–175. Botany 83, 1168–1174. Brown, B.J., Mitchell, R.J., 2001. Competition for pollination: effects Memmott, J., Waser, N.M., 2002. Integration of alien plants into a of pollen of an invasive plant on seed set of a native congener. native flower–pollinator visitation web. Proceedings of the Oecologia 129, 43–49. Royal Society of London Series B-Biological Sciences 269, Brown, B.J., Mitchell, R.J., Graham, S.A., 2002. Competition for 2395–2399. pollination between an invasive species (purple loosestrife) Mitchell, R.J., 1997. Effects of pollination intensity on Lesquerella and a native congener. Ecology 83, 2328–2336. fendleri seed set – variation among plants. Oecologia 109, Chittka, L., Schu¨ rkens, S., 2001. Successful invasion of a floral 382–388. market – an exotic Asian plant has moved in on Europe’s Moragues, E., Traveset, A., 2005. Effect of Carpobrotus spp. on the river-banks by bribing pollinators. Nature 411, 653. pollination success of native plant species of the Balearic Feldman, T.S., Morris, W.F., Wilson, W.G., 2004. When can two Islands. Biological Conservation 122, 611–619. plant species facilitate each other’s pollination? Oikos 105, Palmer, T.M., Stanton, M.L., Young, T.P., 2003. Competition and 197–207. coexistence: exploring mechanisms that restrict and maintain Fishbein, M., Venable, D.L., 1996. Diversity and temporal change in diversity within mutualist guilds. American Naturalist 162, the effective pollinators of Asclepias tuberosa. Ecology 77, S63–S79. 1061–1073. Price, M.V., Waser, N.M., Irwin, R.E., Campbell, D.R., Brody, A.K., Gegear, R.J., Laverty, T.M., 1998. How many flower types can 2005. Temporal and spatial variation in pollination of a bumble bees work at the same time? Canadian Journal of montane herb: A seven-year study. Ecology 86, 2106–2116. Zoology 76, 1358–1365. Rathcke, B., 1983. Competition and facilitation among plants for Gegear, R.J., Laverty, T.M., 2004. Effect of a colour dimorphism on pollination. In: Real, L. (Ed.), Pollination Biology. Academic the flower constancy of honey bees and bumble bees. Press, Inc., Orlando, FL, pp. 305–329. Canadian Journal of Zoology 82, 587–593. Roubik, D.W., 2001. Ups and downs in pollinator populations: Herrera, C.M., 1988. Variation in mutualisms: the spatio-temporal when is there a decline? Conservation Ecology, 5. mosaic of a pollinator assemblage. Biological Journal of the SAS Institute, Inc., 1999. SAS OnlineDoc, Version 8. SAS Institute, Linnean Society 35, 95–125. Inc., Cary, NC. Herrera, C.M., 1995. Microclimate and individual variation in Selleck, G.W., Coupland, R.T., Frankton, C., 1962. Leafy spurge in pollinators: flowering plants are more than their flowers. Saskatchewan. Ecological Monographs 32, 1–29. Ecology 76, 1516–1524. Spira, T.P., 2001. Plant–pollinator interactions: A threatened Johnson, S.D., Steiner, K.E., 2000. Generalization versus mutualism with implications for the ecology and specialization in plant pollination systems. Trends in Ecology management of rare plants. Natural Areas Journal 21, 78–88. and Evolution 15, 140–143. Trammell, M.A., Butler, J.L., 1995. Effects of exotic plants on native Kearns, C.A., Inouye, D.W., 1993. Techniques for Pollination ungulate use of habitat. Journal of Wildlife Management 59, Biologists. University Press of Colorado, Niwot, Colorado. 808–816. Kearns, C.A., Inouye, D.W., 1994. Fly pollination of Linum lewisii Waser, N.M., 1978. Interspecific pollen transfer and competition (Linaceae). American Journal of Botany 81, 1091–1095. between co-occurring plant species. Oecologia 36, 223–236. Kearns, C.A., Inouye, D.W., Waser, N.M., 1998. Endangered Waser, N.M., Chittka, L., Price, M.V., Williams, N.M., Ollerton, J., mutualisms: The conservation of plant–pollinator 1996. Generalization in pollination systems, and why it interactions. Annual Review of Ecology and Systematics 29, matters. Ecology 77, 1043–1060. 83–112.